Concordance between whole-slide imaging and light microscopy for routine surgical pathology

Interesting article from the University of Nebraska, a well-known Department of Pathology with a number of years of digital pathology and whole slide imaging experience under their belts. There is also an editorial by Wally Henricks in Journal of Pathology Informatics (full PDF version).

The paper entitled “Concordance between whole-slide imaging and light microscopy for routine surgical pathology” (Human Pathology subscription required) looked diagnostic concordance between light microscopy and whole slide imaging with 158 consecutive cases (312 parts, 1085 slides). Overall concordance was 96.5%. Of the discordant cases (5), two were felt to have been clinically significant and due to interpretive or diagnostic error, unrelated to image quality, according to the authors, although reference is made to some issues with resolving nuclear detail and microorganisms.

Of note, the authors mention a rescan rate in 13% of cases.

Increasingly, as we discussed in our paper “Whole-slide imaging: Routine Pathologic Diagnosis” (full PDF version) published earlier this year (with many thanks to authors Toby Cornish and Ryan Swapp), we too discovered in our review, high concordance rates between light microsocpy and digital pathology, and despite the somewhat fragmented and flawed nature of individual studies, the sum of this evidence strongly suggests that whole-slide imaging is ready for clinical use.

Increasingly, studies, such as one looking at “Reliability of whole slide images as a diagnostic modality for renal allograft biopsies” and “Superiority of virtual microscopy versus light microscopy in transplantation pathology” confirm the collective experience when put to the test showing that, in these cases, whole-slide imaging may not only be equivalent to glass slides, it may actually be better.

I gather additional studies with selected, consecutive routine work, “routine” cases with an enrichment of “difficult” cases with continue to show high diagnostic concordance when tried and tested with qualified readers.

The question of whether or not whole-slide imaging is equivalent to the microscope then appears to be getting answered and repeatedly, answered in the affirmative.

We could ask ourselves “Is the light microscope inferior to whole-slide imaging” rather than asking if the new technology is at least as good as the old.

Now the conclusions and experiences show, as these authors and Wally point out, that workflow considerations with current technologies, including re-scan needs (perhaps not as high as 13% across the board) to deal with non-trivial matters when reviewing cases.

Arguably, a difficult case is a difficult case, and whole-slide imaging, which is required to create an exact digital reproduction of a glass slide into a digital format that can be recognized by the human eye, should not enhance or degrade the image in any way that is different from the glass slide image.

Albeit, “re-scan” may equal “defer to glass slide” much like a consult is deferred to another pathologist or a frozen section glass slide diagnosis is “deferred to permanent sections”.

This will increasingly, with continuously improving technologies, continue to be the exception rather than the rule.

Publishers note: I say this every year and usually it doesn’t work but this may be the last post for the year with the holiday break upon us.  My many thanks to you the reader and our sponsors for continued support.  I hope this site over the past more than 5 years has been a source of trusted information and news whether your interests be academic, business related or out of curiosity. 

In advance of the end of the world as we know it — 

Happy holidays and best wishes for the new year,

Keith